Robust domain adaptation with noisy and shifted label distribution

Shao-Yuan Li, Shi-Ji Zhao, Zheng-Tao Cao, Sheng-Jun Huang, Songcan Chen

Abstract: Unsupervised Domain Adaptation (UDA) intends to achieve excellent results by transferring knowledge from labeled source domains to unlabeled target domains in which the data or label distribution changes. Previous UDA methods have acquired great success when labels in the source domain are pure. However, even the acquisition of scare clean labels in the source domain needs plenty of costs as well. In the presence of label noise in the source domain, the traditional UDA methods will be seriously degraded as they do not deal with the label noise. In this paper, we propose an approach named Robust Self-training with Label Refinement (RSLR) to address the above issue. RSLR adopts the self-training framework by maintaining a Labeling Network (LNet) on the source domain, which is used to provide confident pseudo-labels to target samples, and a Target-specific Network (TNet) trained by using the pseudo-labeled samples. To combat the effect of label noise, LNet progressively distinguishes and refines the mislabeled source samples. In combination with class rebalancing to combat the label distribution shift issue, RSLR achieves effective performance on extensive benchmark datasets.